From bbc8ee17a859a85a4432a6fe21cf534fc88e751f Mon Sep 17 00:00:00 2001 From: Martin Vignali Date: Sat, 17 Nov 2018 23:48:58 +0100 Subject: [PATCH 12/13] avcodec/proresdec : add 12b decoding based on patch by Kieran Kunhya --- libavcodec/proresdec.h | 8 +++++ libavcodec/proresdec2.c | 65 ++++++++++++++++++++++++++++++++++---- tests/fate/prores.mak | 8 ++--- tests/ref/vsynth/vsynth3-prores | 4 +-- tests/ref/vsynth/vsynth3-prores_ks | 4 +-- 5 files changed, 75 insertions(+), 14 deletions(-) diff --git a/libavcodec/proresdec.h b/libavcodec/proresdec.h index 3d99eba078..80d00964b0 100644 --- a/libavcodec/proresdec.h +++ b/libavcodec/proresdec.h @@ -34,7 +34,14 @@ typedef struct { int ret; } SliceContext; +enum ProresDecodingPrecision { + PRORES_PRECISION_AUTO = 0, + PRORES_PRECISION_10 = 10, + PRORES_PRECISION_12 = 12 +}; + typedef struct { + AVClass *class; BlockDSPContext bdsp; ProresDSPContext prodsp; AVFrame *frame; @@ -50,6 +57,7 @@ typedef struct { const uint8_t *scan; int first_field; int alpha_info; + enum ProresDecodingPrecision decoding_precision; /* user value */ void (*unpack_alpha)(GetBitContext *gb, uint16_t *dst, int num_coeffs, const int num_bits); } ProresContext; diff --git a/libavcodec/proresdec2.c b/libavcodec/proresdec2.c index f819f8db21..bc46f6a47a 100644 --- a/libavcodec/proresdec2.c +++ b/libavcodec/proresdec2.c @@ -29,6 +29,7 @@ #define LONG_BITSTREAM_READER #include "libavutil/internal.h" +#include "libavutil/opt.h" #include "avcodec.h" #include "get_bits.h" #include "idctdsp.h" @@ -155,15 +156,31 @@ static av_cold int decode_init(AVCodecContext *avctx) break; case MKTAG('a','p','4','h'): avctx->profile = FF_PROFILE_PRORES_4444; + avctx->bits_per_raw_sample = 12; break; case MKTAG('a','p','4','x'): avctx->profile = FF_PROFILE_PRORES_XQ; + avctx->bits_per_raw_sample = 12; break; default: avctx->profile = FF_PROFILE_UNKNOWN; av_log(avctx, AV_LOG_WARNING, "Unknown prores profile %d\n", avctx->codec_tag); } + if (ctx->decoding_precision == PRORES_PRECISION_10) { + avctx->bits_per_raw_sample = 10; + av_log(avctx, AV_LOG_DEBUG, "Force 10b decoding"); + } else if (ctx->decoding_precision == PRORES_PRECISION_12) { + avctx->bits_per_raw_sample = 12; + av_log(avctx, AV_LOG_DEBUG, "Force 12b decoding"); + } else {/* auto mode, use codec tag */ + if (avctx->bits_per_raw_sample == 10) { + av_log(avctx, AV_LOG_DEBUG, "Auto bitpdepth precision. Use 10b decoding based on codec tag"); + } else { /* 12b */ + av_log(avctx, AV_LOG_DEBUG, "Auto bitpdepth precision. Use 12b decoding based on codec tag"); + } + } + ff_blockdsp_init(&ctx->bdsp, avctx); ret = ff_proresdsp_init(&ctx->prodsp, avctx); if (ret < 0) { @@ -211,6 +228,7 @@ static int decode_frame_header(ProresContext *ctx, const uint8_t *buf, width = AV_RB16(buf + 8); height = AV_RB16(buf + 10); + if (width != avctx->width || height != avctx->height) { av_log(avctx, AV_LOG_ERROR, "picture resolution change: %dx%d -> %dx%d\n", avctx->width, avctx->height, width, height); @@ -237,9 +255,17 @@ static int decode_frame_header(ProresContext *ctx, const uint8_t *buf, } if (ctx->alpha_info) { - avctx->pix_fmt = (buf[12] & 0xC0) == 0xC0 ? AV_PIX_FMT_YUVA444P10 : AV_PIX_FMT_YUVA422P10; + if (avctx->bits_per_raw_sample == 10) { + avctx->pix_fmt = (buf[12] & 0xC0) == 0xC0 ? AV_PIX_FMT_YUVA444P10 : AV_PIX_FMT_YUVA422P10; + } else { /* 12b */ + avctx->pix_fmt = (buf[12] & 0xC0) == 0xC0 ? AV_PIX_FMT_YUVA444P12 : AV_PIX_FMT_YUVA422P12; + } } else { - avctx->pix_fmt = (buf[12] & 0xC0) == 0xC0 ? AV_PIX_FMT_YUV444P10 : AV_PIX_FMT_YUV422P10; + if (avctx->bits_per_raw_sample == 10) { + avctx->pix_fmt = (buf[12] & 0xC0) == 0xC0 ? AV_PIX_FMT_YUV444P10 : AV_PIX_FMT_YUV422P10; + } else { /* 12b */ + avctx->pix_fmt = (buf[12] & 0xC0) == 0xC0 ? AV_PIX_FMT_YUV444P12 : AV_PIX_FMT_YUV422P12; + } } avctx->color_primaries = buf[14]; @@ -585,6 +611,7 @@ static void decode_slice_alpha(ProresContext *ctx, } block = blocks; + for (i = 0; i < 16; i++) { memcpy(dst, block, 16 * blocks_per_slice * sizeof(*dst)); dst += dst_stride >> 1; @@ -606,6 +633,7 @@ static int decode_slice_thread(AVCodecContext *avctx, void *arg, int jobnr, int LOCAL_ALIGNED_16(int16_t, qmat_chroma_scaled,[64]); int mb_x_shift; int ret; + uint16_t val_no_chroma; slice->ret = -1; //av_log(avctx, AV_LOG_INFO, "slice %d mb width %d mb x %d y %d\n", @@ -643,7 +671,8 @@ static int decode_slice_thread(AVCodecContext *avctx, void *arg, int jobnr, int chroma_stride = pic->linesize[1] << 1; } - if (avctx->pix_fmt == AV_PIX_FMT_YUV444P10 || avctx->pix_fmt == AV_PIX_FMT_YUVA444P10) { + if (avctx->pix_fmt == AV_PIX_FMT_YUV444P10 || avctx->pix_fmt == AV_PIX_FMT_YUVA444P10 || + avctx->pix_fmt == AV_PIX_FMT_YUV444P12 || avctx->pix_fmt == AV_PIX_FMT_YUVA444P12) { mb_x_shift = 5; log2_chroma_blocks_per_mb = 2; } else { @@ -684,18 +713,24 @@ static int decode_slice_thread(AVCodecContext *avctx, void *arg, int jobnr, int else { size_t mb_max_x = slice->mb_count << (mb_x_shift - 1); size_t i, j; + if (avctx->bits_per_raw_sample == 10) { + val_no_chroma = 511; + } else { /*12b */ + val_no_chroma = 511 * 4; + } for (i = 0; i < 16; ++i) for (j = 0; j < mb_max_x; ++j) { - *(uint16_t*)(dest_u + (i * chroma_stride) + (j << 1)) = 511; - *(uint16_t*)(dest_v + (i * chroma_stride) + (j << 1)) = 511; + *(uint16_t*)(dest_u + (i * chroma_stride) + (j << 1)) = val_no_chroma; + *(uint16_t*)(dest_v + (i * chroma_stride) + (j << 1)) = val_no_chroma; } } /* decode alpha plane if available */ - if (ctx->alpha_info && pic->data[3] && a_data_size) + if (ctx->alpha_info && pic->data[3] && a_data_size) { decode_slice_alpha(ctx, (uint16_t*)dest_a, luma_stride, buf + y_data_size + u_data_size + v_data_size, a_data_size, slice->mb_count); + } slice->ret = 0; return 0; @@ -803,6 +838,23 @@ static av_cold int decode_close(AVCodecContext *avctx) return 0; } +#define OFFSET(x) offsetof(ProresContext, x) +#define VD AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_DECODING_PARAM +static const AVOption options[] = { + {"precision", "set decoding precision", OFFSET(decoding_precision), AV_OPT_TYPE_INT, {.i64=0}, PRORES_PRECISION_AUTO, PRORES_PRECISION_12, VD, "precision"}, + {"auto", "use codec tag (10b for 422, 12b for 444)", 0, AV_OPT_TYPE_CONST, {.i64=PRORES_PRECISION_AUTO}, INT_MIN, INT_MAX, VD, "precision"}, + {"10", "force 10b decoding", 0, AV_OPT_TYPE_CONST, {.i64=PRORES_PRECISION_10}, INT_MIN, INT_MAX, VD, "precision"}, + {"12", "force 12b decoding", 0, AV_OPT_TYPE_CONST, {.i64=PRORES_PRECISION_12}, INT_MIN, INT_MAX, VD, "precision"}, + { NULL }, +}; + +static const AVClass prores_class = { + .class_name = "PRORES", + .item_name = av_default_item_name, + .option = options, + .version = LIBAVUTIL_VERSION_INT, +}; + AVCodec ff_prores_decoder = { .name = "prores", .long_name = NULL_IF_CONFIG_SMALL("ProRes (iCodec Pro)"), @@ -815,4 +867,5 @@ AVCodec ff_prores_decoder = { .decode = decode_frame, .capabilities = AV_CODEC_CAP_DR1 | AV_CODEC_CAP_SLICE_THREADS | AV_CODEC_CAP_FRAME_THREADS, .profiles = NULL_IF_CONFIG_SMALL(ff_prores_profiles), + .priv_class = &prores_class, }; diff --git a/tests/fate/prores.mak b/tests/fate/prores.mak index dd52d68f70..b23714b65b 100644 --- a/tests/fate/prores.mak +++ b/tests/fate/prores.mak @@ -15,10 +15,10 @@ fate-prores-422: CMD = framecrc -flags +bitexact -i $(TARGET_SAMPLES)/pror fate-prores-422_hq: CMD = framecrc -flags +bitexact -i $(TARGET_SAMPLES)/prores/Sequence_1-Apple_ProRes_422_HQ.mov -pix_fmt yuv422p10le fate-prores-422_lt: CMD = framecrc -flags +bitexact -i $(TARGET_SAMPLES)/prores/Sequence_1-Apple_ProRes_422_LT.mov -pix_fmt yuv422p10le fate-prores-422_proxy: CMD = framecrc -flags +bitexact -i $(TARGET_SAMPLES)/prores/Sequence_1-Apple_ProRes_422_Proxy.mov -pix_fmt yuv422p10le -fate-prores-alpha: CMD = framecrc -flags +bitexact -i $(TARGET_SAMPLES)/prores/Sequence_1-Apple_ProRes_with_Alpha.mov -pix_fmt yuva444p10le -fate-prores-alpha_skip: CMD = framecrc -flags +bitexact -skip_alpha 1 -i $(TARGET_SAMPLES)/prores/Sequence_1-Apple_ProRes_with_Alpha.mov -pix_fmt yuv444p10le -fate-prores-transparency: CMD = framecrc -flags +bitexact -i $(TARGET_SAMPLES)/prores/prores4444_with_transparency.mov -pix_fmt yuva444p10le -fate-prores-transparency_skip: CMD = framecrc -flags +bitexact -skip_alpha 1 -i $(TARGET_SAMPLES)/prores/prores4444_with_transparency.mov -pix_fmt yuv444p10le +fate-prores-alpha: CMD = framecrc -flags +bitexact -precision 10 -i $(TARGET_SAMPLES)/prores/Sequence_1-Apple_ProRes_with_Alpha.mov -pix_fmt yuva444p10le +fate-prores-alpha_skip: CMD = framecrc -flags +bitexact -precision 10 -skip_alpha 1 -i $(TARGET_SAMPLES)/prores/Sequence_1-Apple_ProRes_with_Alpha.mov -pix_fmt yuv444p10le +fate-prores-transparency: CMD = framecrc -flags +bitexact -precision 10 -i $(TARGET_SAMPLES)/prores/prores4444_with_transparency.mov -pix_fmt yuva444p10le +fate-prores-transparency_skip: CMD = framecrc -flags +bitexact -precision 10 -skip_alpha 1 -i $(TARGET_SAMPLES)/prores/prores4444_with_transparency.mov -pix_fmt yuv444p10le fate-prores-gray: CMD = framecrc -flags +bitexact -c:a aac_fixed -i $(TARGET_SAMPLES)/prores/gray.mov -pix_fmt yuv422p10le #Test bsf prores-metadata diff --git a/tests/ref/vsynth/vsynth3-prores b/tests/ref/vsynth/vsynth3-prores index 2998dc1d79..367d374dc7 100644 --- a/tests/ref/vsynth/vsynth3-prores +++ b/tests/ref/vsynth/vsynth3-prores @@ -1,4 +1,4 @@ 6afd345a8f799d0459229349a30497cd *tests/data/fate/vsynth3-prores.mov 105367 tests/data/fate/vsynth3-prores.mov -fff5e7ad21d78501c8fa4749bf4bf289 *tests/data/fate/vsynth3-prores.out.rawvideo -stddev: 2.80 PSNR: 39.17 MAXDIFF: 27 bytes: 86700/ 86700 +dfd6ca041c31df20f2e7919d6c3a37d1 *tests/data/fate/vsynth3-prores.out.rawvideo +stddev: 49.75 PSNR: 14.19 MAXDIFF: 216 bytes: 86700/ 86700 diff --git a/tests/ref/vsynth/vsynth3-prores_ks b/tests/ref/vsynth/vsynth3-prores_ks index 99cfc1397e..f23a3b65af 100644 --- a/tests/ref/vsynth/vsynth3-prores_ks +++ b/tests/ref/vsynth/vsynth3-prores_ks @@ -1,4 +1,4 @@ 7ceff8c9cffca766f8a167ba73dad0e2 *tests/data/fate/vsynth3-prores_ks.mov 95053 tests/data/fate/vsynth3-prores_ks.mov -9ab6d3e3cc7749796cd9fa984c60d890 *tests/data/fate/vsynth3-prores_ks.out.rawvideo -stddev: 4.09 PSNR: 35.88 MAXDIFF: 35 bytes: 86700/ 86700 +4c9b4a247982be95990b7450a36de31b *tests/data/fate/vsynth3-prores_ks.out.rawvideo +stddev: 49.48 PSNR: 14.24 MAXDIFF: 211 bytes: 86700/ 86700 -- 2.14.3 (Apple Git-98)